The White House’s highly anticipated artificial intelligence and cybersecurity executive order will designate a coalition of national security and civilian agencies to ramp up scrutiny of cutting-edge AI models, five people familiar with the initiative told POLITICO.
Industry officials began receiving details on the directive from the White House on Tuesday night, said four of the people familiar with discussions about the executive order. It could come as early as Thursday, three people knowledgeable about the planning said.
The people were granted anonymity to share details of ongoing, highly confidential discussions.
One key question is whether the order will require a federal review of advanced AI models before they can be released, an idea the Trump administration had distanced itself from earlier this month after initially floating the proposal. The latest draft would ask developers of the systems to submit certain models to a voluntary review by a phalanx of federal agencies as far as 90 days before they’re made public, people familiar with it said.
The expected executive action represents the administration’s latest step to tighten scrutiny over developers of advanced AI systems and address the risk of catastrophic harm while maintaining a broadly pro-innovation approach to the technology. The move comes amid a policy clash between Anthropic and the Pentagon over the company’s efforts to limit certain military uses of its technology, as well as the subsequent release of Anthropic’s powerful Mythos model to a select group of companies. Weeks of conflicting signals from the administration about the evolving policy had left some people following the policy perplexed about where the White House was heading.
“Everybody’s involved,” one person familiar with the matter said of the order. “That’s why it’s been on the table and off the table and on the table and off.”
The draft executive order is expected to contain at least two sections. The first focuses on cybersecurity and the second on “covered frontier models,” three of the people said. The first segment gives the Pentagon 30 days to secure its networks, including key telecommunications and information systems.
The administration would have 30 days to mandate wider AI use across government systems and critical infrastructure organizations, such as community banks, rural hospitals and utilities, according to three of the people.
The directive also tasks the Treasury Department with leading voluntary work with AI industry officials to form a clearinghouse within 30 days, according to five of the people familiar with details of the order. The effort will establish a voluntary partnership with AI-makers and critical infrastructure owners and operators to find and patch vulnerabilities, the people said. Other federal agencies including the Office of the National Cyber Director, the National Security Agency and the Cybersecurity and Infrastructure Security Agency will support the effort, according to the people.
The Office of Personnel Management will be tasked with increasing hiring at the U.S. Tech Force, a program announced by OPM Director Scott Kupor late last year to attract top AI talent to various federal agencies, according to three of the people familiar with the order.
The second section is expected to give a cohort of federal agencies and offices — including Treasury, CISA and the National Institute of Standards and Technology — 60 days to establish a classified benchmarking process to determine what constitutes a “covered frontier model” under the executive order. White House chief of staff Susie Wiles, National Cyber Director Sean Cairncross and Michael Kratsios — who leads the White House Office of Science and Technology Policy — would also be involved in the benchmarking process.
The NSA would ultimately be tasked with making final determinations in consultation with other agencies.
The executive order will ask AI developers participating in the voluntary framework to do three things: engage with the government before releasing models covered by the order, give the government access to those models 90 days before public release, and share access to select critical infrastructure ahead of launch.
Some details of the draft executive order, including the 90-day timeline for voluntary review, were previously reported by Axios.
Spokespeople from ONCD, NIST, Treasury and the NSA did not immediately respond to requests for comment. A spokesperson for the Department of Homeland Security, which houses CISA, deferred to the White House, which did not immediately respond to a request for comment.The directive comes as the rollout of highly advanced, cyber-focused models such as Mythos has rattled governments and key sectors worldwide.
Shortly after Anthropic unveiled Mythos to a limited audience of tech companies and security researchers last month, the White House began meeting with the tech and cyber industry to discuss potential details on the directive. But the drafting process has revealed fractures within the Trump administration over the best path to secure the frontier models.
Maggie Miller contributed to this report.
